Communities and Culture: Findings and Research Needs
The studies reviewed here suggest that some neighborhood characteristics related to disorganization and instability (heterogeneity, residential instability, family disruption, collective efficacy) are associated with both violent and nonviolent crime and the pattern is not consistent with the differential etiology of violence. We did find that analyses of “disorder” (social or physical) are more consistently associated with violent crime than they are with nonviolent crime, but, given the small number of studies, we categorize this as inconclusive (see Table 12.1). Importantly, indicators of social disorganization remain consistently associated with violence even controlling for concentrated disadvantage; studies that apply this control have reported effects of social disorganization on violent crime that exceed the effects on nonviolent crime. The fact that high violent crime rates and high property crime rates happen in the same communities is not surprising and does not prove that community factors have equivalent effects on both types of crime. While this set of studies provides evidence leaning in the direction of the differential etiology of violence, it suffers from a lack of direct tests similar to those seen in other chapters such as comparisons of violent and nonviolent offenders. Importantly, testing whether the magnitude of effects on violence are larger than the effects on nonviolent crime would fill some gaps in the literature, and testing whether community factors predict violent crime, controlling for nonviolent crime, would fill a yawning chasm in our knowledge base. Existing data sets could easily be brought to bear for this purpose.
The qualitative studies also indicate nuanced ways that collective efficacy may be suppressed by the presence of violence, a finding that is worthy of its own line of research. We also feel that the weight of qualitative evidence indicates that a subculture of violent values is a key factor responsible for notably high rates of violence sustained in certain disadvantaged communities. The quantitative studies of “subcultural” theories are few, and only one in this review included an indicator of nonviolent offending to provide any evidence about this violence-specific theory, but the qualitative evidence is strong and researchers have emphasized violent subcultures in many works. Unfortunately, contrasts between descriptions of neighborhoods in the work of Anderson (1999) and Patillo (1998) are limited by selection biases, and only one study has focused on an area of low violence and high crime. However, the existence of any such places stands in favor of the hypothesis that subcultural values can probably distinguish between neighborhoods of high and low violence, high and low drug dealing, high and low property crime, etc. Isolating the effects of culture would require comparing neighborhoods similar in socioeconomic factors which vary in measurable ways on indicators of violent subculture. This particular line of research is likely to have more theoretical than practical interest.
An important issue in the literature on communities and crime is the use of multiple indicators of the same construct (usually social disorganization) in the same models. This makes the interpretation of findings tenuous. For example, if an indicator of residential instability, such as population change, is included in the same model with a measure of “migrants" divorce rates, percent Black and percent Hispanic, interpreting the coefficient for “migrants” becomes very narrow. Authors should carefully consider their research question before specifying multivariate models. If that question is narrowly focused on which indicators of disorganization are most important, or whether migrants (not disadvantaged minorities) are an important indicator of social disorganization, then this type of conservative modeling may be appropriate. In studies of social disorganization, in our view, this is an example of model “overspecification,” and in many studies, effects may have been missed or underestimated.